Forced instability of core-annular flow in capillary constrictions Igor Beresnev, William Gaul,

advertisement
PHYSICS OF FLUIDS 23, 072105 (2011)
Forced instability of core-annular flow in capillary constrictions
Igor Beresnev,1 William Gaul,2 and R. Dennis Vigil2
1
Department of Geological and Atmospheric Sciences, Iowa State University, 253 Science I,
Ames, Iowa 50011-3212, USA
2
Department of Chemical and Biological Engineering, Iowa State University, 2114 Sweeney Hall,
Ames, Iowa 50011-2230, USA
(Received 20 January 2011; accepted 14 June 2011; published online 22 July 2011)
Instability of fluid cylinders and jets, a highly nonlinear hydrodynamic phenomenon, has fascinated
researchers for nearly 150 years. A subset of the phenomenon is the core-annular flow, in which a
non-wetting core fluid and a surrounding wall-wetting annulus flow through a solid channel. The
model, for example, represents the flow of oil in petroleum reservoirs. The flow may be forced to
break up when passing through a channel’s constriction. Although it has long been observed that
the breakup occurs near the neck of the constriction, the exact conditions for the occurrence of the
forced breakup and its dynamic theory have not been understood. Here, we test a simple geometric
conjecture that the fluid will always break in the constrictions of all channels with sufficiently long
wavelengths, regardless of the fluid properties. We also test a theory of the phenomenon. Four
constricted glass tubes were fabricated above and below the critical wavelength required for
the fluid disintegration. In a direct laboratory experiment, the breakup occurred according to the
conjecture: the fluids were continuous in the shorter tubes but disintegrated in the longer tubes. The
evolution of the interface to its pinch-off was recorded using high-speed digital photography.
The experimentally observed core-annulus interface profiles agreed well with the theory, although
the total durations of the process agreed less satisfactorily. Nonetheless, as the theory predicts, the
ratio between the experimental and theoretical times of the breakup process tends to one with
decreasing capillary number. The breakup condition and the dynamic theory of fluid disintegration
in constricted channels can serve as quantitative models of this important natural and technical
C 2011 American Institute of Physics. [doi:10.1063/1.3607472]
phenomenon. V
I. INTRODUCTION
Spontaneous breakup of liquid cylinders and jets into
isolated drops is a fascinating natural phenomenon that has
captured researchers’ attention since the time of Lord Rayleigh [Ref. 1, Secs. 357–359]. The effect is known as the
Plateau-Rayleigh instability. A related phenomenon is
instability of the core-annular flow, or the flow of two fluids in a solid capillary channel, in which one (the core)
occupies the center of the channel and the other forms the
surrounding wall-wetting film. Both phenomena are freesurface, capillarity-controlled flows, a class of strongly
nonlinear problems for which no exact theories exist [Ref.
2, Secs. 2–3 and 4–3].
A typical approximate theoretical approach to understanding the onset of fluid instability is to subject an initially cylindrical free interface to a small sinusoidal
disturbance and determine the wavelength that has the fastest growth rate, assuming the linearity in the early stages of
evolution. This is the subject of the linear stability analysis.
The size of the resulting drops for both the free jets and the
core-annular flows in straight cylindrical capillaries is controlled by this critical wavelength, which depends on the
viscosities and the radii of the core and the annulus [Ref. 1,
Secs. 357–359; Refs. 3–6]. The linear-stability approach
provides a useful practical means to determine under what
conditions the flow is stable and under what it is not; how1070-6631/2011/23(7)/072105/7/$30.00
ever, this does not provide a description of the nonlinear
dynamics of the interface during its evolution to a complete
breakup nor does it reveal the characteristic time scale of
the process.
The class of problems discussed so far is the spontaneous breakup in cylindrical channels. For the core-annular
flow, the disintegration process can be forced by driving the
fluids through a constriction. It is known from experience
that, in this case, the core pinches off in the vicinity of the
neck of the constriction. A wavy wall is essential for the observation of the forced-breakup phenomenon. The effect is
widely observed in natural core-annular flows, for example,
in the movement of oil surrounded by water in petroleum
reservoirs. Constricted pore geometry is typical of natural
core-annular flows, in which the potential occurrence of fluid
breakup is no longer spontaneous but imposed by the geometry of the channel. It is believed that the phenomenon
explains why oil predominantly exists in nature in the form
of dispersed tiny droplets.
Although it is known that the forced breakup always
occurs near the necks of the constrictions, it has been all but
enigmatic as to why in some cases it does take place but not
in the others.1–7 A unifying framework capable of explaining
miscellaneous computational and experimental observations,
for arbitrary fluid viscosities, has been lacking. Is it possible
to explain the full gamut of observations from a single theoretical standpoint?
23, 072105-1
C 2011 American Institute of Physics
V
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://phf.aip.org/phf/copyright.jsp
072105-2
Beresnev, Gaul, and Vigil
Phys. Fluids 23, 072105 (2011)
We addressed this question theoretically and computationally in two recent studies. First, starting from an earlier
suggestion that the channel geometry plays an important role
in the occurrence of fluid breakup,13 we derived an approximate pressure distribution in the core phase based on Laplace’s law of capillary pressure.14 The criterion for the
development of instability in a wavy channel can then be formulated as the exceedance in the pressure in the neck of the
profile relative to the pressure in the “crest,” causing the core
fluid to leave the neck. In geometric terms, for a sinusoidal
channel, this leads to an inequality
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
L > 2p rmin rmax ;
(1)
where L is the wavelength of the channel and rmin and rmax
are the minimum and maximum radii of the core phase. It
turned out that all published experimental and computational
studies of the breakup in flows dominated by capillary forces
(that is, having relatively small capillary numbers Ca) in sinusoidal channels, as reviewed by Beresnev et al.,14 complied with this criterion. It is noteworthy that the criterion (1)
is derived from a simple pressure argument, although it
appears to be universally valid. It should, therefore, be more
appropriately called a breakup conjecture. It declares that,
regardless of the viscosities of the fluids, geometry is the
only factor controlling whether the fluid disintegrates or not.
The first goal of the present publication is to subject this conjecture to a direct experimental test.
Second, the geometric condition (1) provides a “yes”/“no”
answer but does not establish a dynamic model of the phenomenon. We consequently developed a hydrodynamic theory of
the forced instability, which, unlike the previous theoretical
studies, is able to follow the interface evolution to a complete
disintegration for arbitrary viscosities lc and lf of the core and
film fluids, respectively (Ref. 15; this publication also summarizes the earlier theoretical studies of the phenomenon). An analytical treatment of this complex hydrodynamic problem is
only possible if one assumes smallness of the slope of the interface. For example, suppose the non-dimensional shape of the
sinusoidal channel in cylindrical coordinates is
kðxÞ ¼ 1 að1 þ cos paxÞ;
(2)
where kðxÞ is the radial position of the channel’s wall, x is
the axial coordinate, a is the geometric parameter quantifying the minimum radius of the pore, and a is the slope constant defining the tube’s wavelength L, a ¼ 2=L (here all
quantities having the dimension of distance have been normalized by the maximum radius of the sinusoidal tube Rt).
The smallness of the slope is then equivalent to the condition
a 1. The small-slope assumption, also supposing the
smallness of capillary and Reynolds numbers, allows the use
of expressions for Poiseuillean flow for the annulus and the
core. The smaller a, the better the approximation. The analysis then leads to a fourth-order nonlinear partial-derivative
equation for the temporal evolution of the non-dimensional
interface profile jðx; sÞ:15
2
3
2lc k2
1þ
1 7
@j
1 lf Q @ 6
lf j2
6
7
¼
6
7
5
@s
2p lc j @x 4 k4 lf
þ
1
4
j
lc
1 @
@j
@3j
j2
þ j2 3
þ
4j @x
@x
@x
99
8 2
>
>
>
k
lc 1 k2 3 lc
>
>
>
>>
>
þ < j2 1
lc k==
lf 4 j2 4 lf
;
ln
>
>
lf j>
k4 lf
>
>
>
>
>
>
þ 1
;;
:
j4 lc
(3)
where Q is the total volume flux of both fluids through the
channel, s is the non-dimensional time (the time normalized
by the characteristic time scale lc Rt =r), and r is the corefilm interfacial tension. The equation reflects the complexity
of the underlying nonlinear physics but allows inexpensive
numerical analysis. An experimental test of the predicted
evolution of the interface to its snap-off, governed by Eq.
(3), is the second goal of the present study.
We are therefore set to perform two tests of the current
theoretical understanding of the forced instability in constricted channels. We test the validity of the geometric conjecture (1) for the occurrence of core-fluid disintegration.
We also validate the correctness of the dynamic theory
expressed by differential Eq. (3). We next describe the laboratory experiments and their results.
II. EXPERIMENT
Four axisymmetric constricted tubes were manufactured
from commercially available straight glass tubing by a glassblower. They will be referred to as the “Short1,” “Short2,”
“Medium,” and “Long” models. The geometry was prescribed in such a way that the wavelengths of the Short1 and
Short2 tubes did not satisfy the condition (1); the core fluid,
therefore, was not expected to snap off. These were the
shortest that the glassblower could produce. On the other
hand, the wavelengths of the Medium and Long tubes satisfied the condition. These tubes differed in the value of their
slope constant: a ¼ 0.36 and 0.061, respectively.
The photographed tube-wall profiles with least-squares fitted sinusoidal lines are shown in Figure 1 (the non-dimensionalization is by Rt). For technical reasons, it was much easier to
manufacture longer-wavelength constricted tubes close to sinusoidal shapes than the shorter ones. Because of this, the Medium
and Long tubes follow nearly perfect sinusoidal lines, while the
short ones (especially Short1) exhibit significant departure.
However, this does not affect our ability to test the breakup criterion, in which only the maximum and minimum radii matter.
Table I summarizes the geometric characteristics of all
tubes. The radii are the optically measured values. In all
cases except Short1, the wavelengths were obtained from the
sinusoidal fitting. For the Short1 tube, the wavelength was
measured as the distance from the center of the constriction
to the point where the radius reached its maximum value.
An experimental apparatus was built to observe the displacement of the suspending fluid by an invading core phase
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://phf.aip.org/phf/copyright.jsp
072105-3
Forced instability of core-annular flow
Phys. Fluids 23, 072105 (2011)
FIG. 1. Profiles of the experimental tubes fitted to sinusoidal shapes.
(Figure 2). The capillary was placed inside a transparent viewing box. The region between the viewing cell and the tube
was filled with glycerol to reduce optical distortion due to the
curvature of the tube. Small ports at either end of the viewing
cell provided access for feeding and removing the working
fluids into or out of the capillary. At the inlet side, there are
two ports: one for the annular fluid and one, covered with a
rubber septum, for the injection of the core fluid. Images of
the tube were captured by a Photron FASTCAM APX-RS
high-speed digital camera mounted above the flow cell.
Aqueous solutions of glycerol were used as suspending
fluids (Table II). The core fluids were trichloroethylene
(TCE)-heptane mixtures dyed with Oil Blue N. The densities
of the two phases were matched to eliminate buoyancy
TABLE I. Geometric parameters of the constricted tubes.
Tube
type
Short1
Short2
Medium
Long
Minimum
radius (mm)
Maximum
radius (mm)
Wavelength
(mm)
Predicted minimum
wavelength for the
breakup (mm)
0.5
0.6
0.4
0.3
4.3
4.6
4.3
4.2
9.2
9.0
24.0
138.5
9.2
10.4
8.2
7.0
effects. The fluid-fluid interfacial tension measured with a
DuNouy tensiometer was 8 103 N=m for all three glycerol=TCE-heptane systems studied.
To remove any organic residue and increase the hydrophilicity of the glass, the capillary was first cleaned in a
NaOH-ethanol-deionized-water solution and then flushed with
deionized water and dried. In a given run, the capillary was
first filled with the annular fluid and the core phase was
injected upstream of the test section via a syringe. The volume
of the injected core droplet was kept constant at 200 ll. A syringe pump then pumped the annular fluid through the capillary and recording of images took place. The recording ended
when the core fluid broke up or the trailing meniscus exited
the constriction. The drop velocity (used to calculate the capillary number) could be controlled by changing the speed of the
syringe pump. The measured velocities are listed in Table III.
The measurement of the annular-film thickness in pixels
was performed in a digital-image editor after the completion
of the experiment. This thickness was converted to absolute
units through the known optical magnification and the
known camera-sensor’s absolute pixel size. Optical resolution of the camera was about 16 lm for the Short and Medium tubes and 1.5 lm for the Long tube. The upstream film
thickness was constant before the snap-off process (if any)
occurred (e.g., see the movie attached as an electronic
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://phf.aip.org/phf/copyright.jsp
072105-4
Beresnev, Gaul, and Vigil
Phys. Fluids 23, 072105 (2011)
FIG. 2. (Color online) Schematic of the
experimental apparatus.
supplement to the article. The units on the axes in the movie
are mm). This constant thickness was taken as the initial condition in the respective theoretical simulation. The time to
breakup was measured by counting the image frames from
the high speed camera for shorter breakup times or by a hand
timer for longer (greater than 60 s) times.
In an experiment with set parameters (tube type, fluid
system, and capillary number), several breakup events were
typically analyzed. If the breakup times were relatively short
and the core drop was long enough, it could undergo several
separate breakup events (if any) as it passed through the constriction. All of them were recorded. For longer breakup
times, only one breakup event usually occurred. In that case,
several experimental runs were performed. The total number
of events analyzed is listed in Table III, in which the film
thicknesses and experimental breakup times are the averages
of all breakup events for given experiment.
III. RESULTS
core phase. The Short1 tube therefore was not expected to
produce breakup if that was exactly the case. In the experimental runs with larger capillary numbers, the film was visible, which led to the reduction in the values of rmin and rmax
relative to the calculation. The breakup condition, therefore,
turned out to be satisfied. At smaller capillary numbers, the
film could no longer be seen, making the assumption of negligible thickness applicable. The breakup was not observed
either.
For the Short2 tube, as calculation using Eq. (1) shows,
assuming a constant film thickness, the film would have to
be thicker than approximately 0.15 mm to reduce the rmin
and rmax for the core phase to make the wavelength of 9.0
mm exceed the threshold. Such thicknesses, even if the film
was visible, were never observed (also see Table III),
explaining why the snap-off never took place in the Short2
tube.
The experiments thus confirmed the occurrence of fluid
instability according to the geometric criterion.
A. Test of the geometric conjecture
B. Validation of the dynamic theory
Table I lists the minimum wavelength of the channel, calculated from inequality [Eq. (1)], above which the core fluid
is conjectured to always pinch off in the neck. The wavelengths of the Medium and Long tubes are above the threshold
value, and the breakup of the core was observed for all capillary numbers (typically between 0.001 and 0.8), except for
one case for the Medium tube discussed in Sec. III B. The
wavelength of the Short2 tube is below the threshold; breakup
in this tube was not observed.
The manufactured wavelength of the Short1 tube turned
out to be exactly at the threshold value of 9.2 mm. The core
fluid snapped off at capillary numbers above approximately
0.03 and did not at their lower values. This behavior can be
understood. The calculation of the threshold wavelengths in
Table I assumed negligible thickness of the wall-wetting
film, or the equality between the radii of the tube and the
TABLE II. Properties of the wetting (annulus) fluid.
Annulus fluid
Glycerol 100% wt
Glycerol 85% wt
Glycerol 75% wt
Viscosity (Pa s)
Density (g=cm3)
1.4
0.11
0.036
1.3
1.2
1.2
In cases of the Medium and Long tubes, we compared
the observed dynamic behavior with that predicted by the
evolution Equation (3), including both the ability of the
theory to predict the shape of the interface and the timing of
the process.
There are two time scales in the process: a long one during most of the thickening of the suspending-fluid lens in the
neck of the constriction and a much shorter one during the
precipitous collapse of the interface at the final stage.11,15
Our earlier numerical simulations showed that this latter
stage is the one where, over the short time scale, most of the
change in the interface profile takes place.15 This final stage,
therefore, provides the crucial testing of the correctness of
the theory.
The summary of experiments is presented in Table III.
The viscosity ratios tested varied by a factor of 40. The values of h are the initial film thicknesses formed by the invasion of the core through the neck of the constriction. The
breakup times (tT - theoretical, tE - experimental) were
counted from the moment when the film was formed until
the completion of the pinch-off.
Figures 3 and 4 present the observed and theoretically
predicted shapes of the core-annulus interface (the flow is to
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://phf.aip.org/phf/copyright.jsp
072105-5
Forced instability of core-annular flow
Phys. Fluids 23, 072105 (2011)
TABLE III. Summary of experiments and breakup-time observations for the Medium and Long tubes.
Medium tube (a ¼ 0.36)
lf=lc ¼ 2820
Number of eventsa ¼ 5
h ¼ 0.11 mm
Drop velocity ¼ 1.2 mm=s
Ca ¼ 0.2
tT ¼ 5.1 s
tE ¼ 10.7 s
tT=tE ¼ 0.48
Long tube (a ¼ 0.061)
103
1
0.096
4.1
0.06
0.7
6.6
0.11
71
0
Unresolved
4.1
0.02
No breakup
2820
4
0.099
4.4
0.8
62
15
4.1
2820
5
0.064
1.3
0.2
332
83
4.0
218
4
0.044
4.4
0.06
84
74
1.1
218
4
0.023
1.3
0.02
497
144
3.5
71
3
0.017
1.3
0.006
316
230
1.4
a
Total number of observed breakup events.
the right). There are differences in the recording of the process between the Medium and Long tubes that are important
for understanding the way the results are shown. As seen
from Table III, the breakup times are much shorter for the
Medium tube (on the order of seconds) compared with the
Long tube (minutes). Because of the precipitous nature of
the collapse of the interface at the final stage, and in order to
capture the dynamics near the final point, the sampling interval of the digital camera had to be sufficiently small. The
limited camera memory then imposed restrictions on the record duration. The entire evolution of the interface from
invasion to pinch-off could still be recorded for the Medium
tube; however, this was not possible for the Long tube
because of the much longer duration. In the latter case, we
could only film a fraction of the total time of the process
near the pinch-off moment.
Figure 3 shows the position of the wall channel as well
as the theoretical and experimental interface profiles for the
Medium tube for two values of the capillary number. The
FIG. 3. Experimental and theoretical interface profiles near breakup for
the Medium tube (enhanced online) [URL: http://dx.doi.org/10.1063/
1.3607472.1].
“snapshots” were taken very near collapse. Numerical
restrictions did not allow advancing the theoretical interface
arbitrarily close to the channel axis, as a singularity was
quickly approached: j tends to zero in the denominator of
1=j4 in Eq. (3) and the solution “blows up.” The theoretical
and experimental shapes agree very closely.
Figure 3(B) also shows the interface obtained by using
the commercial computational-fluid-dynamics (CFD) code
15
FLUENT. As in the simulations by Beresnev and Deng, FLUENT
was run in the volume of fluid (VOF) axisymmetric model for
the immiscible multi-phase flow. The mesh-generation software GAMBIT was used to construct the grid on which the axial
and radial momentum, continuity, and volume-fraction equations were solved. The following solution schemes for the
VOF model were applied: the PREssure STaggering Option
(PRESTO) for pressure interpolation, the second-order discretization for the volume-fraction equation, the Pressure Implicit
with Splitting of Operators (PISO) scheme for pressurevelocity coupling, and the second-order upwind discretization
for the momentum equations. As seen from Figure 3(B), the
deviation from the experiment around the neck of the constriction in the FLUENT simulation is greater than that for the
theoretical curve obtained from the solution of Eq. (3).
The electronic-supplement movie shows the full
recorded process of the breakup in the experiment corresponding to Figure 3(B), where the complete details of the
evolution can be seen.
In one instance, indicated in Table III, the breakup in
the Medium tube did not take place (Ca ¼ 0.02). This is the
only experiment with the Medium and Long tubes at which
the film was not visible either. Our interpretation of the fact
is that, although the breakup was allowed by the geometry,
the influx of the suspending fluid into the neck was insufficient, due to the thinness of the film, to complete the process
before the drop exited the constriction.
Figure 4 presents the results of the comparison for the
Long tube. The thin lines show the wall and the theoretical
interface near the breakup. The camera’s field of view in the
Long-tube experiment covered a smaller portion of the total
wavelength of the tube: the respective sections of the wall
and the observed interface are shown by the thick lines. For
the reason explained, we filmed only the final stage of the
evolution near the interface collapse, while numerical instability did not allow advancing the theoretical solution arbitrarily close to the collapse. The lower thick line shows the
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://phf.aip.org/phf/copyright.jsp
072105-6
Beresnev, Gaul, and Vigil
experimentally observed initial position of the interface
when the filming started. There is certain mismatch in the
timing at which the theoretical and experimental snapshots
are taken; given this circumstance, the agreement between
the experimental and theoretical shapes can be considered
good. The locations of the pinch-off, slightly shifted downstream from the neck, match better between the theory and
the experiment as the capillary number decreases.
The theory, therefore, is capable of adequately capturing
the details of the shape of the evolving interface for both the
Medium and Long tubes. We also tested the ability of the
theory to explain the correct timing of the process. The ratios
of the theoretical-to-observed breakup times tT=tE are listed
in Table III. For the Medium tube, there is underprediction
by the factors of two and nine. For the Long tube, there is,
conversely, overprediction; however, it tends to decrease
from approximately a factor of four to close to one from
large to small capillary numbers. The greater mismatch
between the theory and experiment for the Medium tube and
for larger capillary numbers is to be expected, recalling the
assumptions of the small slope and small capillary number
built into the theory. Curiously, the ratio of tF=tE, where tF is
the breakup time obtained by FLUENT, corresponding to the
CFD simulation shown in Figure 3(B), is 1.1. FLUENT, thus,
FIG. 4. Experimental and theoretical interface profiles near breakup for the
Long tube.
Phys. Fluids 23, 072105 (2011)
better than the theory reproduced the timing of the process
but not the shape of the interface.
IV. CONCLUSIONS
Instability and disintegration of the core-annular flow
can be forced by passing the fluids through a capillary constriction. Such a forced breakup has been observed in some
instances but not in the others, with the deciding factors
remaining unclear.
We formulated a “geometric breakup conjecture” stating
that, regardless of the fluid properties, the channel geometry
is the factor determining whether the fluid snap-off will take
place. The geometric condition (1) is derived from simple
pressure argument and is consistent with the literature on the
phenomenon. The criterion (1) is counterintuitive at first
glance: it indicates that the fluid in longer channels with
gently sloping walls is susceptible to instability, whereas that
in “sharper” channels with short wavelengths will flow continuously. The explanation of this phenomenon will bear on
our ability to understand the state of oil in tortuous pore
spaces in petroleum reservoirs.
Four constricted glass tubes were fabricated below and
above the critical wavelength defined by inequality (1). In all
tubes, the breakup occurred as expected from the conjecture.
A theory of the evolution of the interface to its disintegration, for arbitrary fluid viscosities, has not been known either. We have proposed the nonlinear evolution Equation (3)
as a quantitative model of the process. The smaller the slope
parameter a and the capillary number, the more accurate the
theory is expected to be. The interface profiles and breakup
times, following from this equation, were compared to the
experimentally observed ones in the tubes with two slope parameters, a ¼ 0.36 and 0.061. For both tubes, the experimental shapes of the interface near the breakup agreed with the
theoretical prediction, the match improving with decreasing
capillary number. Note that one consequence of the imposed
flow is the occurrence of the breakup downstream of the constriction, as seen in Figures 3 and 4; the exact location of the
pinch-off is also theoretically captured better as Ca
decreases. The agreement in the timing of the process is less
satisfactory. Nonetheless, the ratio between the theoretical
and experimental breakup times for the smallest-slope tube
tends to one with decreasing capillary number, which also is
an expectation of the theory.
Note that we have not attempted to extend the applicability of both the geometric conjecture and the dynamic
evolution equation into the range of capillary numbers
exceeding one. The derivation of the equation and the formulation of the criterion explicitly assume the preponderance of
capillary forces over the viscous ones, expressed in the
smallness of Ca. Our experiments thus stayed approximately
within the limits of applicability of the approach, to which
the conclusions apply. However, in the capillary-number
range tested, from 0.006 to 0.8, there is still a variety of
observed behaviors (Table III). In the practical applications
to oil recovery that we have already mentioned and in the
flow of fluids through natural porous media, the smallness of
the capillary number is a very good approximation.16
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://phf.aip.org/phf/copyright.jsp
072105-7
Forced instability of core-annular flow
The theory and the yes=no criterion for the occurrence
of forced breakup in core-annular flow in capillary constrictions compare favorably against experiment. They can serve
as quantitative models of this widespread natural and technical phenomenon within their limits of applicability.
ACKNOWLEDGMENTS
This study was supported through National Science
Foundation (Award No. EAR-0602556) and Petroleum
Research Fund (Award No. 45169-AC9). The CFD simulations for Figure 3 were performed by W. Deng. The authors
are indebted to Robert Ewing for helpful discussions and to
the two anonymous referees whose comments improved the
clarity of the presentation.
1
J. W. S. Rayleigh, The Theory of Sound (Dover, New York, 1945), Vol. II.
S. Middleman, Modeling Axisymmetric Flows (Academic, New York,
1995).
3
S. Tomotika, “On the instability of a cylindrical thread of a viscous liquid
surrounded by another viscous fluid,” Proc. R. Soc. London, Ser. A 150,
322 (1935).
4
S. L. Goren, “The instability of an annular thread of fluid,” J. Fluid Mech.
12, 309 (1962).
5
H. H. Hu and D. D. Joseph, “Lubricated pipelines: Stability of core-annular flow. Part 2,” J. Fluid Mech. 205, 359 (1989).
2
Phys. Fluids 23, 072105 (2011)
6
K. Preziosi, K. Chen, and D. D. Joseph, “Lubricated pipelines: Stability of
flow,” J. Fluid Mech. 201, 323 (1989).
W. L. Olbricht and L. G. Leal, “The creeping motion of immiscible drops
through a converging=diverging tube,” J. Fluid Mech. 134, 329 (1983).
8
T. M. Tsai and M. J. Miksis, “Dynamics of a drop in a constricted capillary
tube,” J. Fluid Mech. 274, 197 (1994).
9
M. Hemmat and A. Borhan, “Buoyancy-driven motion of drops and bubbles in a periodically constricted capillary,” Chem. Eng. Commun. 148150, 363 (1996).
10
M. J. Martinez and K. S. Udell, “Axisymmetric creeping motion of drops
through a periodically constricted tube,” in Drops and Bubbles, edited by
T. G. Wang (American Institute of Physics, New York), AIP Conf. Proc.
197, 222 (1989).
11
P. A. Gauglitz and C. J. Radke, “The dynamics of liquid film breakup in
constricted cylindrical capillaries,” J. Colloid Interface Sci. 134, 14
(1990).
12
C. Kouris and J. Tsamopoulos, “Core-annular flow in a periodically constricted circular tube. Part 2. Nonlinear dynamics,” J. Fluid Mech. 470,
181 (2002).
13
J. G. Roof, “Snap-off of oil droplets in water-wet pores.” Soc. Pet. Eng. J.
10, 85 (1970).
14
I. A. Beresnev, W. Li, and R. D. Vigil, “Condition for break-up of nonwetting fluids in sinusoidally constricted capillary channels,” Transp. Porous Media 80, 581 (2009).
15
I. A. Beresnev and W. Deng, “Theory of breakup of core fluids surrounded
by a wetting annulus in sinusoidally constricted capillary channels,” Phys.
Fluids 22, 012105-1 (2010).
16
J. C. Melrose and C. F. Brandner, “Role of capillary forces in determining
microscopic displacement efficiency for oil recovery by waterflooding,” J.
Can. Pet. Technol. 13, 54 (1974).
7
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://phf.aip.org/phf/copyright.jsp
Download